ES2395966A1 - System and method for the remote management of a supercomputing platform using a web interface - Google Patents

Abstract

The invention relates to a system and method for the remote management of a supercomputing platform using a web interface, by means of a user interface (100) via a web browser (502) of a client station (504). The method comprises: dynamically generating a computation initialisation form (112) for the introduction of parameters, initial conditions and contour conditions of the problem, including obtaining the concrete model (200) of the problem to be solved, loading an XML metadata structure (202) associated with said model, using an advanced XSL style template (204) to transform (206) the XML structure, and obtaining the parameters, initial conditions and contour conditions of the problem for the subsequent execution in the cluster (508) selecting execution parameters (120) for executing the processes to be executed in the cluster (508) including data and compiling (122) the resulting code according to the execution parameters selected by the user and executing (126) the resulting code in computing nodes of the cluster (514).

Description

System and method of remote management of a supercomputing platform via web interface.

Field of the invention The present invention falls within the field of supercomputing platforms used especially for the numerical simulation of problems described by mathematical models related to physics, statistics, engineering, biology, economics, etc. that require a large computing capacity and are used in multidisciplinary environments such as simulation of geophysical flows. In this area it has been applied for the modeling of floods in rural areas, simulations of marine flows of one or two superimposed layers of fluids, simulation of subarea and submarine avalanches, simulation of tsunamis, simulation of sediment transport and turbid currents, etc.

BACKGROUND OF THE INVENTION Solving numerical simulation problems is usually performed using a computer (or a cluster of computers - a cluster is a group of multiple computers linked by a high-speed network -) in which certain pre-installed software is used. and adapted to the computer (or computer cluster) itself. Thus, for example, for the simulation of geophysical flows there are currently specific software packages. This software is usually limited to a certain operating system and certain hardware specifications. Among the programs that solve geophysical flow problems are Mike 21 (DHI software), Fluent (Ansys), Comsol Multiphysics (Comsol) and Telemac 2D (EDF). These require the installation of specific programs or the export of graphic terminals for the interaction between the user and the software.

When it comes to solving real large problems, the power of a personal computer is usually insufficient, making it necessary to use a supercomputing cluster. Not all existing software is adapted to this type of supercomputing architectures and those that require it require precise hardware specifications, specific installation processes and may require qualified personnel simply for installation.

In addition, most of this type of applications allow at most to execute existing and pre-compiled codes, but usually do not allow the inclusion of source code by the user or the possibility of compilation for execution in a supercomputing cluster. In those that allow the inclusion of source code by the user, they are oriented to the resolution of specific problems and the interaction with the application is usually done through console-type terminals, exporting graphic terminals or by installing specific software on the user equipment final.

One of the most complete programs in the field of simulation of geophysical flows is Fluent, which allows the definition of specific functions by the user (UDF) that are compiled before the execution process. However, this software requires installation not only in the supercomputing cluster, but also in specific programs on the user's computer both for the interaction with the Fluent computer core and for the handling of pre and postprocessing tools.

The cluster is usually in a situation that is not accessible to the user (special rooms for laboratories where the computer equipment and restricted access is physically located). Interacting with this type of software usually requires tools for remote access and export of graphic terminals, which require the intervention of qualified personnel that enable the end user to use said software package. Applications that allow remote management of a cluster through a graphical interface are known. These applications require installation on a client computer. Also known are applications that allow monitoring of a cluster using a simple web browser. Through these applications the user can check the load of the cluster and even display certain results.

Among the scientific literature, the document “A web-based platform for computer simulation of seismic ground response” (Zhaohui et al, Journal Advances in Engineering Software. Volume 35, Issue 5, May 2004 Elsevier Science Ltd. Oxford, UK) presents a web platform for simulation of seismic effects. The input interface is implemented through a web browser. The input data can be entered in a static form, which is prepared exclusively for the problem to be solved, and by entering a file. During the execution of the simulation, the user is informed of the status of the simulation, for example through a web page. As output interface describes a dynamic web page where you can view the results and manipulate them. However, the web platform that it discloses is designed exclusively for the resolution of a specific problem, it is not applicable to different types of problems, so that for any other embodiment everything would have to be redesigned from the beginning. Another related scientific document is entitled “Simulation anywhere any time: web-based simulation implementation for evaluating order-to delivery system and progress” (Kumara et al; Proceedings of the Winter Simulation Conference, 2002 volume 2. 11/12/2002. ISBN: 0-7803-7614-5). This document describes a three-tier client / server simulation framework based on the internet. The main elements are a database server, a runtime server and a web-based user interface. The web-based graphical user interface helps users easily perform the simulation study from anywhere and at any time, and also makes it easier to view the results.

Document US20080178179-A1 describes a system and method for a set of middleware components to support the execution of financial risk analysis applications on a supercomputing platform. The user sends the jobs from an application or a web browser and the client uses an XML-based file to send the jobs.

Document US2008082933-A1 discloses a graphical user interface that allows access to the user account and show the status of the work. An input panel allows you to specify the input data file. The graphical interface can be implemented through an interactive web server accessible through standard web browser technology. Once the user accesses their account, they can select the virtual power center (VPC or supercomputer cluster). The graphical interface shows the list of applications available in the VPC, their current status and usage statistics. The graphical user interface allows the user to choose the number of processors for the job execution.

There are also different documents in which the user accesses the supercomputing cluster for the management, submission of works or obtaining results through standard web browsers, such as documents WO2005106664, US2003028645 and the article "Web Access to Supercomputing" (Alosio et al .; Computing in Science & Engineering December 2001. Volume 3, Issue 6. Pages 66-72. ISSN 1521-9615).

Although some of the previous inventions employ a web interface as an element of interaction between user and supercomputing cluster, however, none allows the dynamic generation of preprocess forms for each mathematical model with the possibility of including source code by the user. The codes that describe the aforementioned documents are already compiled and are executed based on the parameters entered in the entry forms, so the user is not allowed to enter their own source code to later be compiled together with the existing codes. Likewise, in the above-mentioned inventions the calculations must be executed in conventional CPUs and are not prepared to be executed in GPUs (graphics card processors).

Description of the invention The present invention solves the aforementioned problems, by creating a cluster platform and web interface that provides the user with the ability to numerically solve problems described by mathematical models. The web interface frees the user from the dependence of installing software on the client computer at the cost of a much greater integration between said interface and the cluster hardware that, together with advanced web programming, enables dynamic user-code interaction in an environment friendly similar to other commercial programs but under a purely web interface.

The supercomputing platform of the present invention is composed of a cluster of computers interconnected with each other and a new user interface made entirely with web technology. This type of platform is intended for multidisciplinary companies and researchers interested in the numerical resolution of problems described through mathematical models.

A first aspect of the invention relates to a method of remote management of a supercomputing platform by means of a web interface, said remote management of the computer cluster being carried out through a user interface implemented by means of a web browser of a client station. The method comprises:

-

 a dynamic generation stage of a computation initialization form for the introduction by the user of the parameters, initial and contour conditions of the problem to be solved, where said generation stage comprises:

• obtain the specific model of the problem to be solved, according to the user's selection;

• load an XML structure of code metadata associated with the selected model;

• use an XSL advanced style template to transform the XML structure and show the user the computation initialization form associated with the selected model;

• obtain the parameters, initial conditions and contour of the user's problem for subsequent execution in the cluster;

-

 a step of selecting the execution parameters of the processes to be executed in the computer cluster;

-

 a stage of data inclusion and compilation, whereby the computer cluster includes the data of the initialization form of computations in the code of the selected model and compiles the resulting code based on the execution parameters selected by the user;

-

 an execution stage of the resulting code in compute nodes of the cluster according to the execution parameters selected by the user.

The data inclusion and compilation stage preferably comprises: obtaining resources necessary for compilation, including:

• XML template associated with the model that indicates the necessary input data;

•

 Modifiable source code template with initial condition and contour data and associated with the model selected by the user;

•

 XML file with the data entered by the user in the computation initialization form;

• precompiled library associated with the model selected by the user;

-

 With the resources obtained, generate a code initialization file and generate the custom source code with the data entered by the user in the code initialization form.

The step of selecting the execution parameters may comprise the selection of compute nodes of the cluster for execution. You can also understand the selection of one of the following execution modes:

-

 sequential or parallel;

-

 in CPU or GPU architecture.

The code execution stage may comprise:

-

 check the execution mode selected by the user;

-

 select the hardware resources of the cluster necessary for execution, checking their availability;

-

 Once the hardware resources are available, carry out the code execution.

The code execution stage may also comprise updating the cluster database with the execution data to allow the user to monitor the status of the execution by means of cluster monitoring tools and the running processes included in the user interface.

The method may include sending the computer initialization form filled out by the user to the computer cluster for error verification and validation.

Another aspect of the present invention relates to a remote management system of a supercomputing platform via web interface, comprising:

-

a cluster of computers with an access server, a front-end of the cluster and a plurality of compute nodes in the cluster; Y

-

 a user interface implemented through a web browser of a client station.

The access server is configured to obtain the specific problem model to solve according to the user's selection. Said access server comprises:

-

 a form management module responsible for the dynamic generation of a form for initialization of computations for the introduction by the user of the parameters, initial conditions and contour of the problem to be solved, by:

•

 loads an XML structure of code metadata associated with the selected model;

•

the use of an advanced XSL style template to transform the XML structure and send it to the web browser of the client station so that the computation initialization form associated with the selected model is displayed to the user;

-

 An execution module responsible for:

• receive the execution parameters of the processes to be executed in the cluster of computers selected by the user through the web browser on the client station;

• manage the execution of processes with the front-end of the cluster.

The front end of the cluster is configured to:

•

include the data of the initialization form of computations in the code of the selected model and compile the resulting code based on the execution parameters selected by the user;

•

run the resulting code on the cluster compute nodes according to the parameters

of execution selected by the user. The server is preferably configured to obtain and forward to the front end of the cluster resources necessary for compilation, including:

• XML template associated with the model that indicates the necessary input data;

•

 Modifiable source code template with initial condition and contour data and associated with the model selected by the user;

•

 XML file with the data entered by the user in the computation initialization form;

In the previous case, the front-end of the cluster preferably has access to a precompiled library associated with the model selected by the user and is configured to, with the resources obtained, generate a code initialization file and generate the custom source code with the data entered by the user in the code initialization form.

Execution parameters may include the selection of cluster compute nodes for execution and / or the selection of one of the following execution modes:

-

 sequential or parallel;

-

 in CPU or GPU architecture.

For code execution, the front end of the cluster can be additionally configured to:

-

 check the execution mode selected by the user;

-

 select the hardware resources of the cluster necessary for execution, checking their availability;

-

 Once the hardware resources are available, carry out the code execution.

The supercomputing platform allows easy access to these modules directly from a web browser without the need to install any specific software. In addition, the calculations made through this platform are carried out directly in a computer cluster.

The present invention makes available to users a large computing capacity far superior to that which would be obtained with a simple home computer. The present invention allows the user to access, through a multiplatform user interface based on web technology, a supercomputing cluster in an easy and intuitive, fast and transparent way for the user.

This computing platform brings a number of advantages:

•

Independence of the operating system of the user's computer.

•

Use of universal standards (web technology).

•

Automatic software update without user intervention.

•

Use of a supercomputing cluster from a personal computer.

•

The user does not need to perform installation, maintenance, software and / or hardware update tasks or specific supercomputing knowledge.

•

Transparent and intuitive use regardless of specific user training. It provides a valid multidisciplinary tool for researchers in the scientific field (physics, geology, applied mathematics, ...), environmental consultants, coordination centers for the prevention of natural disasters, etc.

The present invention allows the user:

•

Interact with the cluster as if it were working locally.

•

Avoid installing any specific software on your local computer.

•

Take advantage of the parallel and supercomputing architecture of the cluster (both CPU and GPU). The user is allowed to manage the topology of a parallel machine (selecting the necessary cores and starting or stopping the parallel machine at his convenience) in the event that his simulation requires the use of multiple computing units.

•

Possibility of modifying existing codes according to your needs that are automatically compiled at runtime for the CPU or GPU architecture selected by the user. The compilation process incorporates a compilation status control system.

•

Control at all times the processes that are running allowing you to stop running calculations at all times, even when they are running on hundreds of nodes.

•

Have information on available resources: CPU load, memory and available compute nodes, etc.

You can also optionally provide other advantages:

•

Authentication using LDAP and federated SSH.

•

Advanced file manager via web.

•

Possibility of incorporating specific pre and post process tools via the web for each problem:

o Preprocess:

•

Mesh generators.

•

Interpolation modules for initial condition data.

•

Information extraction modules from available scientific databases (they are usually used for the generation of topographies, bathymetries, data readings for the imposition of initial or boundary conditions, etc.).

•

Etc.

o Postprocess:

•

1D, 2D and 3D analytical results visualizers.

•

Converters of file types.

•

Etc.

BRIEF DESCRIPTION OF THE FIGURES Next, a series of figures that help to better understand the invention and that expressly relate to an embodiment of said invention which is presented as a non-limiting example thereof is described very briefly.

Figure 1 shows the flowchart of the supercomputing platform.

Figure 2 shows an operation scheme of the initialization form of computations.

Figure 3 shows the code compilation process scheme.

Figure 4 shows the scheme for launching executions in the cluster.

Figure 5 shows a possible scheme of the supercomputing platform.

DETAILED DESCRIPTION OF THE INVENTION Figure 1 shows the flow chart of the supercomputing platform, with its various elements:

-

 User interface 100, with the following characteristics:

*

The user accesses the supercomputing platform through a standard web browser.

*

 In a particular embodiment the authentication of the users of the academic network is carried out by means of the RedIRIS federation system.

* In the event that the user has administrator rights, it also allows:

-

 Management of other users through LDAP (Lightweight Directory Access Protocol) systems.

-

 Permission management of each user: codes to which he has access, nodes of the cluster in which he can launch executions, etc.

-

 Cluster hardware management: physical characteristics of each node (CPU, memory, GPU, etc.).

-

 Management of codes managed by the supercomputing platform.

Initially, the user interface 100 is shown where the user authenticates on the platform to access a personalized menu where the different options are accessed. The user can then use the following options:

-

Preprocessing tools 102 (optional).

-

Postprocessing tools 104 (optional).

-

Management tools for your file system 106.

-

Cluster monitoring tools and running processes 108.

-

Administration tools (in the case of administrators) 110.

-

Computation initialization form 112.

One of the essential elements for the execution of problems in the supercomputing platform is the initialization form of computations 112 (its generation process is described below). This form is fully adapted for each problem and through it the user enters the parameters, initial conditions and contour for each problem to be solved (which optionally have been pretreated with preprocessing tools 102). As represented by the circular arrow, the form adapts dynamically according to the data entered by the user. Once the user completes this form, he sends it for verification to cluster 114 where each field is analyzed according to the defined type, verifying that the information provided matches the specifications of said type 116. In the case of errors, it is returned to show the form to the user to correct them. If the verification is correct, the platform assigns the form a unique identifier and is saved, then taking the user to the user interface 100.

Once the user has completed a computation initialization form, the next step is to launch the numerical simulation in the computing cluster. To do this, the user must initially select from the forms that he has completed 118.

Subsequently, the user must select process execution parameters 120. In this step the platform can display:

-

Real-time monitoring of the computational load of each node in the cluster: percentage of use of each CPU, number of cores used of each CPU, available and used memory in each node, etc.

-

User-accessible compute nodes and execution priority in them.

-

Automatic creation of the calculation topology for parallel processes.

-

Possibility of personalized selection by the user of the compute nodes for a parallel calculation.

The user selects the node (s) he wishes and orders the execution. The cluster includes the data of the initialization form of computations in the code of the selected model, compiles the resulting code 122 (optionally if a parallel computation is necessary, the platform starts a parallel machine with the topology selected by the user) and verifies the result of compilation 124. In case of any compilation error, the errors found are returned to the user and the user is taken to interface 100. If the compilation is correct, the platform executes in the cluster (in the mode selected by the user -sequential, parallel or GPU-) the resulting code 126. Until the results 128 are obtained, the user has tools to track his executions 108 (estimated remaining time, display of trace files, control of executions, etc. ), that is, a self-management of processes in execution and completion time allowing user control through a web interface.

One of the cores of the supercomputing platform is the design of the computation initialization forms 112. Figure 2 shows an operation scheme of the computation initialization form. These forms are the interface between the user and each code and are the tool for the initialization of each scientific calculation that the end user wishes to launch.

The following are summarized in the following items which are the fundamental elements of these forms:

1. Design of a specific tag metalanguage to differentiate between the different possible input types in a code that solves numerical simulation problems. Incorporation of an intrinsic error detection system to this metalanguage.

The metalanguage developed consists of two modules: a module based on XML type language and a complementary one based on XSL style sheet languages.

A sophisticated structured system has been developed under an XML format to organize and store all the types of data needed for a numerical simulation code. This system includes elements linked to the types of usual fields (text fields, various options selectors, etc.) in a web form and advanced elements adapted for the introduction of special data for a numerical simulation problem (for example, for the introduction of boundary conditions).

Each element has specific attributes as well as a set of associated and necessary child elements for the process of entering, validating and verifying the data entered.

The following describes, by way of example, two elements of this metalanguage, one of the basic type and the other advanced.

Integer element:

o Use: Storage of fields of type whole numbers.

o Attributes:

•

name: field name.

•

required (yes / no): indicates if it is a field that the user must necessarily complete.

o Associated child elements:

•

label: contains the text shown to the user as a description of the text field.

•

min: Optional element. Indicates the minimum possible value to enter.

•

max: Optional element. Indicates the maximum possible value to enter.

•

fieldcomment: Optional element. Indicates help text that is displayed to the user in the form.

•

value: Optional element. Value of this field.

•

example: Optional element. Show examples to the user.

o Example:

<integer name = "nodosmpi" required = "yes">

<label> Nodes to be used in MPI parallelization: </label>

 <min> 0 </min>

 <max> 48 </max>

<fieldcomment> You must enter the number of machine nodes

parallel </fieldcomment>

<value />

</integer>

Contour element:

o Use: Imposition of the boundary conditions associated with the problem considered.

o Attributes:

•

name: field name.

•

filename: indicates the name of the field in which the name of the mesh file on which the code is to be executed is stored. This attribute is associated with an advanced function that allows you to dynamically examine the borders of the mesh, reading it from the user's account and dynamically displaying the borders of the mesh classified by references.

•

required (yes / no): indicates if it is a field that the user must necessarily complete.

o Associated child elements:

• condition: Used to describe the possible boundary conditions that the user can select for a given border. This child node will be dynamically replicated as many times as there are borders in the selected mesh.

•

 Attributes:

o name: field name.

o image (yes / no): indicates whether or not the figures associated with the meshing borders should be displayed.

•

Associated child elements:

o label: Text shown to the user.

o minselecteditems: Optional element. Minimum number of items to select.

o maxselecteditems: Optional element. Maximum number of items to select.

• options: used to store the possible boundary condition options at each border.

• Associated child elements:

• Option: Text corresponding to the option.

o Example:

<contour name = "cc_ref" required = "no" filename = "ficheromallado">

<condition name = "ccontorno" image = "yes"> <label> Type of CC for reference </label> <minselecteditems> 0 </minselecteditems> <maxselecteditems> 2 </maxselecteditems>

<options> <option> Elevation </option> <option> Q * n </option>

</options>  <value /> </condition> </contour>

2. Automated system for generating specific forms from templates designed with this metalanguage. In these forms the data related to the problem to be solved is introduced, such as initial conditions, boundary conditions, parameters, etc.

3.

 Possibility of completing empty forms or recovering data from a pre-existing one.

Four.

 Special fields that allow:

•

Select a file stored remotely in the user account of the cluster by accessing its file system in real time.

•

Incorporate the contents of a local or remote file directly into the form.

•

Generation of images from data entered by the user. They are interesting for example for the imposition of boundary conditions graphically at the edges of the spatial domain of the problem. For this, the mesh or spatial domain selected by the user is dynamically interpreted.

•

Fields that evolve dynamically from the user's input data: repetition of structures, conditionals, multiple selection, etc.

•

Specific error control by fields.

Figure 2 shows the dynamic process of generating and validating forms for the initialization of computations (steps 112, 114 and 116 of Figure 1). The flow chart is described below:

Once the user selects 200 the model with which he wants to perform his numerical experiment, the supercomputing platform loads the XML structure associated with said code 202, and uses the XSL 204 advanced style template (this template is common to all forms) to transform the XML structure and show the user the computation initialization form associated with the code that you want to execute 206. This form can also incorporate unique features:

•

Direct access to the user's account in the cluster to complete different types of fields 208 (meshing files, initial condition files, etc.).

•

Specific dynamic fields 210 for the imposition of boundary conditions (fields that determine the boundaries of the selected mesh and automatically represent the boundaries classified by references for the user to indicate the type of boundary conditions at each border) and special data entry structures (For example, fields that replicate data structures to adapt to the incorporation of new variables to the starting problem, or fields that modify in form automatically based on the response that the user has given to a given question).

The last point of the flowchart shows the process of verification 114 and validation 116 of the data entered. For this, the supercomputing platform verifies that the mandatory fields have been filled in and also the consistency of the data entered (for example, verifies that the paths to files are correct, that the numerical values are within the specified limits, that the types of data match the specified types, etc.).

As for the process of data inclusion and compilation 122 of Figure 1, Figure 3 shows the code compilation process scheme.

-

With the data entered by the user, the platform automatically generates optimal executables for the specific architecture of the selected nodes. This entire process is automatic, performed in the cluster remotely without the need for installation of any type of library or dependencies on the client machine and regardless of the operating system used by the user.

-

At the same time, error control is performed at compile time.

The flowchart of Figure 3 is described below:

The supercomputing platform collects all the necessary resources 300 for compilation:

-

XML template associated with model 302 that indicates the necessary input data.

-

Template of modifiable source code 304 associated with the model selected by the user.

-

XML 118 file corresponding to the information provided by the user in the selected form.

-

Precompiled library 308 associated with the model selected by the user.

With this information the supercomputing platform generates an initialization file of code 310 and creates the customized source code with the data entered by user 312 in the code initialization form.

The files generated are compiled against the code's own library following the hardware specifications indicated by the user in the code initialization form and according to the hardware of the cluster in which the user has execution permits 314. The compilation process incorporates a Dynamic error control. In this case, the compilation errors are collected and displayed in an exit form to the user.

Figure 4 shows the execution launch scheme in the supercomputing platform

126. The flowchart is described below:

The code (compiled as shown in the process of data inclusion and compilation 122) will be executed. For this, 400 the specifications given by the user in the code initialization form are analyzed. In particular, it is studied whether the execution is sequential or in parallel and in CPU or GPU architecture. The platform then prepares 402 and checks 404 for the availability of the hardware resources (CPUs and / or GPUs) of the platform necessary for execution. In case the hardware resources are not available (because they are being used) the platform queues the system processes execution. Once the resources are available, execution 406 is carried out. In said process, the code is registered in the database of platform 408 so that the user can monitor at all times:

-

The status of your active executions.

-

Traceability of the execution status.

-

Possibility of interruption of running processes.

-

Notice by e-mail of the completion of the process.

In a preferred embodiment the internal operation of the cluster is based on the Linux operating system, equipped with specific tools for modeling and numerical simulation.

* Hardware: Supercomputing cluster based on Intel architecture composed of a Frontend machine, calculation nodes with CPU and GPU architecture and SAN type machines.

* Software (Backend):

-

 GNU and specific compilers for Intel architecture.

-

 Libraries of numerical calculation.

-

 Parallel calculation libraries.

-

 Systems for monitoring and tracking the status of each node in the cluster, jobs, etc ... based on Ganglia.

-

 Queue management system: Sun Grid Engine (SGE).

-

 Python

* Software (Frontend):

 -

PHP

-

 AJAX

 -

XML, XSLT.

-

CSS.

-

 Python

 -

 MySQL

 -

O3D

Figure 5 represents, by way of example, a scheme with the hardware of the supercomputing platform. It shows the user 500 accessing through a web browser 502 of a terminal or client station 504 (a personal computer, a tablet, a PDA, etc.), via Internet 506, to a cluster of computers 508. In One possible embodiment can be accessed through the access server 510 of the supercomputing platform, which is responsible for receiving the information from the client station 504, interacting with the front end of the cluster 512 and coordinating tasks between the compute nodes of cluster 514 connected in a local high-speed network 516.

Claims (2)

1- Remote management method of a supercomputing platform through a web interface, said remote management of the computer cluster (508) being carried out through a user interface (100) implemented through a web browser (502) of a client station (504) ), characterized in that it comprises:

-

 a dynamic generation stage of a computation initialization form (112) for the introduction by the user (500) of the parameters, initial and contour conditions of the problem to be solved, where said generation stage comprises:

• obtain the specific model (200) of the problem to be solved, according to the user's selection (500);

•

 load an XML structure of code metadata (202) associated with the selected model;

•

 use an advanced XSL style template (204) to transform (206) the XML structure and show the user the computation initialization form associated with the selected model;

• obtain the parameters, initial and contour conditions of the user's problem (500) for subsequent execution in the cluster (508);

-

 a step of selecting the execution parameters (120) of the processes to be executed in the computer cluster (508);

-

 a stage of data inclusion and compilation (122), whereby the computer cluster (508) includes the data of the initialization form in the code of the selected model and compiles the resulting code based on the parameters of execution selected by the user;

-

 an execution stage (126) of the code resulting in compute nodes of the cluster (514) according to the execution parameters selected by the user.

2- Method according to claim 1, characterized in that the step of data inclusion and compilation (122) includes: - obtaining resources (300) necessary for compilation, including: • XML template associated with the model (302) that indicates the necessary input data;

•

source code template (304) modifiable with initial condition and contour data and associated with the model selected by the user;

•

XML file (118) with the data entered by the user in the computation initialization form (112);

• precompiled library (308) associated with the model selected by the user;

-

 with the resources (300) obtained, generate a code initialization file (310) and generate the custom source code (312) with the data entered by the user in the code initialization form (112).

3- Method according to any of the preceding claims, characterized in that the step of selecting the execution parameters (120) comprises the selection of compute nodes of the cluster (514) for execution. 4- Method according to any of the preceding claims, characterized in that the step of selecting the execution parameters (120) comprises the selection of one of the following execution modes:

-

 sequential or parallel;

-

 in CPU or GPU architecture.

5- Method according to any of the preceding claims, characterized in that the execution step (126) of the code comprises:

-

 check (400) the execution mode selected by the user;

-

 select the resources (402) hardware of the cluster necessary for the execution, checking its availability (404);

-

 Once the hardware resources are available, carry out the execution (406) of the code.

6- Method according to any of the preceding claims, characterized in that the execution stage (126) of the code comprises updating the database of the cluster (408) with the execution data to allow the user to monitor the status of the execution by means of tools for monitoring the cluster and the processes in execution (108) included in the user interface (100). 7- Method according to any of the preceding claims, characterized in that it comprises sending to the computer cluster (508) the computation initialization form filled out by the user for the verification of errors (114) and the validation (116) thereof. 8- Remote management system of a supercomputing platform through a web interface, comprising:

-

 a cluster of computers (508) with an access server (510), a front end of the cluster

(512) and a plurality of cluster compute nodes (514); Y

-

 a user interface (100) implemented by web browser (502) of a client station (504); characterized in that the access server (510) is configured to obtain the specific model

(200) of problem to be solved according to the user's selection (500), said access server (510) comprising:

-

 a form management module responsible for the dynamic generation of a form of initialization of computations (112) for the introduction by the user (500) of the parameters, initial conditions and contour of the problem to be solved, by:

•

 loads an XML code metadata structure (202) associated with the selected model;

•

 the use of an advanced style template XSL (204) to transform (206) the XML structure and send it to the web browser (502) of the client station (504) so that the computation initialization form associated with the selected model is shown to the user (500);

-

 An execution module responsible for:

• receive the execution parameters of the processes to be executed in the computer cluster (508) selected by the user through the web browser (502) on the client station (504); • manage the execution of processes with the front-end of the cluster (512); and why the front end of the cluster (512) is configured to:

•

include the data of the initialization form of computations in the code of the selected model and compile the resulting code based on the execution parameters selected by the user;

•

execute the resulting code in the cluster compute nodes (514) according to the execution parameters selected by the user.

9. System according to claim 8, characterized in that the server (510) is configured to obtain and refer to the front end of the cluster (512) resources (300) necessary for compilation, including: • XML template associated with the model (302) that indicates the necessary input data; • source code template (304) modifiable with initial and boundary condition data and associated with the model selected by the user; • XML file (118) with the data entered by the user in the form initialization of computations (112); and because the front end of the cluster (512) has access to a precompiled library (308) associated with the model selected by the user and is configured to generate, with the resources (300) obtained, generate a code initialization file ( 310) and generate the custom source code (312) with the data entered by the user in the code initialization form (112). 10. System according to any of claims 8 to 9, characterized in that the execution parameters include the selection of cluster compute nodes (514) for execution. 11- System according to any of claims 8 to 10, characterized in that the execution parameters include the selection of one of the following execution modes:

-

 sequential or parallel;

-

 in CPU or GPU architecture.

12- System according to any of claims 8 to 11, characterized in that for execution (126) of the code the front end of the cluster (512) is additionally configured to:

-

 check (400) the execution mode selected by the user;

-

 select the resources (402) hardware of the cluster necessary for the execution, checking its availability (404);

-

 Once the hardware resources are available, carry out the execution (406) of the code.

SPANISH OFFICE OF THE PATENTS AND BRAND Application no .: 201230897 SPAIN Date of submission of the application: 08.06.2012 Priority Date: REPORT ON THE STATE OF THE TECHNIQUE 51 Int. Cl.: G06F9 / 50 (2006.01) RELEVANT DOCUMENTS

Category

56 Documents cited Claims Affected

X

US 2008178179 A1 (MITSUMORI S .; NATARAJAN R.; PHAN T.) 24-7-2008, 1-12

TO

ZHAOHUI YANG, JINCHI LU, AHMED ELGAMAL "A web-based platform for computer simulation of seismic ground response"; Advances in Engineering Software 35; Pp 249-259; Available at: http://www.quakesim.net/cyclic1d/web_based_platform_yang_etal.pdf 1, 5-7

TO

US 2008082933 A1 (MASSIVELY PARALLEL TECHNOLOGIES) 3-4-2008, 1.3-7

Category of the documents cited X: of particular relevance Y: of particular relevance combined with other / s of the same category A: reflects the state of the art O: refers to unwritten disclosure P: published between the priority date and the date of priority submission of the application E: previous document, but published after the date of submission of the application

This report has been prepared • for all claims • for claims no:

Date of realization of the report 16.01.2013

Examiner M. Muñoz Sanchez Page 1/4

REPORT OF THE STATE OF THE TECHNIQUE Application number: 201230897 Minimum documentation searched (classification system followed by classification symbols) G06F Electronic databases consulted during the search (name of the database and, if possible, terms of search used) INVENES, EPODOC, WPI State of the Art Report Page 2/4  WRITTEN OPINION Application number: 201230897 Date of Completion of Written Opinion: 01.01.2013 Statement

Novelty (Art. 6.1 LP 11/1986)

Claims Claims 1-12 IF NOT

Inventive activity (Art. 8.1 LP11 / 1986)

Claims Claims 1-12 IF NOT

The application is considered to comply with the industrial application requirement. This requirement was evaluated during the formal and technical examination phase of the application (Article 31.2 Law 11/1986).  Opinion Base.- This opinion has been made on the basis of the patent application as published. State of the Art Report Page 3/4  WRITTEN OPINION Application number: 201230897  1. Documents considered.- The documents belonging to the state of the art taken into consideration for the realization of this opinion are listed below.

Document

Publication or Identification Number publication date

D01

US 2008178179 A1 (MITSUMORI S .; NATARAJAN R.; PHAN T.) 24.07.2008

D02

Zhaohui Yang, Jinchi Lu, Ahmed Elgamal "A web-based platform for computer simulation of seismic ground response"; Advances in Engineering Software 35; Pp 249-259; Available at: http://www.quakesim.net/cyclic1d/web_based_platform_yang_etal.pdf " 04/13/2004

D03

US 2008082933 A1 (MASSIVELY PARALLEL TECHNOLOGIES) 04.04.2008

 2. Statement motivated according to articles 29.6 and 29.7 of the Regulations for the execution of Law 11/1986, of March 20, on Patents on novelty and inventive activity; quotes and explanations in support of this statement D01 is considered the closest document of the state of the art to the object of the request.  Independent claims Claim 1: Document D01 discloses a computer system with a supercomputing platform that uses a web interface and an XML-based file to send jobs that will pass through a middleware layer in charge of managing execution in the supercomputing platform ( parallel computing). The content of an XML file, as is well known and therefore evident to the person skilled in the art, comprises data structures (templates) that facilitate the configuration of compute execution tasks. In the case of scientific calculation, the fields of an XML file could represent, and would be natural, initial, boundary conditions and other parameters (document D02) or parameters related to the execution of the computation (document D03) that could be used on the one hand in the compilation of the scientific model itself and on the other hand to control the execution. The difference of claim 1 with document D01 is the use of an XSL advanced style template to show the user initialization form in another format. This characteristic is not considered to be of a technical nature because it is a way of presenting the information as indicated by the Patent Law and therefore document D01 affects the inventive activity of claim 1 according to article 8.1 of the Patent Law. Claim 8: The content of this system claim corresponds directly to that of the method claim 1, indicating only the necessary functionality of the elements and which follows from the operations performed in the method. Thus, document D01 affects the inventive activity of claim 8 according to article 8.1 of the Patent Law.  Dependent claims Claim 2: The content of this claim details the content of claim 1 explaining what was implied therein. Claims 3 and 4: The content of these claims refers to parameters referred to in the analysis of claim 1. In particular some of them are also found in D03. Claim 5: the user input data is checked in D01. In the framework of the verification of the state of the computer system and its monitoring, it is also commonly known to evaluate the hardware resources and manage the work according to the free capacity, not counting if there is none. This fact is illustrated with document D03. Claims 6-7: As stated for the preceding claims, monitoring is common and the use of databases is also reflected in D01 and D03. As for the submission of the initialization form, it would be implicit in the communication with the computing platform. Claims 9-12: The content of these claims corresponds directly to that of the method dependent claims indicating only the necessary functionality of the elements that follows from the operations performed in the method. In conclusion, document D01 affects the inventive activity of claims 2-7, 9-12 according to article 8.1 of the Patent Law. State of the Art Report Page 4/4